Skip to content

Metrics

This document lists all the monitoring metrics provided by Kube-OVN.

ovn-monitor

OVN status metrics:

Type Metric Description
Gauge kube_ovn_ovn_status OVN Health Status. The values are: (2) for standby or follower, (1) for active or leader, (0) for unhealthy.
Gauge kube_ovn_failed_req_count The number of failed requests to OVN stack.
Gauge kube_ovn_log_file_size The size of a log file associated with an OVN component.
Gauge kube_ovn_db_file_size The size of a database file associated with an OVN component.
Gauge kube_ovn_chassis_info Whether the OVN chassis is up (1) or down (0), together with additional information about the chassis.
Gauge kube_ovn_db_status The status of OVN NB/SB DB, (1) for healthy, (0) for unhealthy.
Gauge kube_ovn_logical_switch_info The information about OVN logical switch. This metric is always up (1).
Gauge kube_ovn_logical_switch_external_id Provides the external IDs and values associated with OVN logical switches. This metric is always up (1).
Gauge kube_ovn_logical_switch_port_binding Provides the association between a logical switch and a logical switch port. This metric is always up (1).
Gauge kube_ovn_logical_switch_tunnel_key The value of the tunnel key associated with the logical switch.
Gauge kube_ovn_logical_switch_ports_num The number of logical switch ports connected to the OVN logical switch.
Gauge kube_ovn_logical_switch_port_info The information about OVN logical switch port. This metric is always up (1).
Gauge kube_ovn_logical_switch_port_tunnel_key The value of the tunnel key associated with the logical switch port.
Gauge kube_ovn_cluster_enabled Is OVN clustering enabled (1) or not (0).
Gauge kube_ovn_cluster_role A metric with a constant '1' value labeled by server role.
Gauge kube_ovn_cluster_status A metric with a constant '1' value labeled by server status.
Gauge kube_ovn_cluster_term The current raft term known by this server.
Gauge kube_ovn_cluster_leader_self Is this server consider itself a leader (1) or not (0).
Gauge kube_ovn_cluster_vote_self Is this server voted itself as a leader (1) or not (0).
Gauge kube_ovn_cluster_election_timer The current election timer value.
Gauge kube_ovn_cluster_log_not_committed The number of log entries not yet committed by this server.
Gauge kube_ovn_cluster_log_not_applied The number of log entries not yet applied by this server.
Gauge kube_ovn_cluster_log_index_start The log entry index start value associated with this server.
Gauge kube_ovn_cluster_log_index_next The log entry index next value associated with this server.
Gauge kube_ovn_cluster_inbound_connections_total The total number of inbound connections to the server.
Gauge kube_ovn_cluster_outbound_connections_total The total number of outbound connections from the server.
Gauge kube_ovn_cluster_inbound_connections_error_total The total number of failed inbound connections to the server.
Gauge kube_ovn_cluster_outbound_connections_error_total The total number of failed outbound connections from the server.

ovs-monitor

ovsdb and vswitchd status metrics:

Type Metric Description
Gauge ovs_status OVS Health Status. The values are: health(1), unhealthy(0).
Gauge ovs_info This metric provides basic information about OVS. It is always set to 1.
Gauge failed_req_count The number of failed requests to OVS stack.
Gauge log_file_size The size of a log file associated with an OVS component.
Gauge db_file_size The size of a database file associated with an OVS component.
Gauge datapath Represents an existing datapath. This metrics is always 1.
Gauge dp_total Represents total number of datapaths on the system.
Gauge dp_if Represents an existing datapath interface. This metrics is always 1.
Gauge dp_if_total Represents the number of ports connected to the datapath.
Gauge dp_flows_total The number of flows in a datapath.
Gauge dp_flows_lookup_hit The number of incoming packets in a datapath matching existing flows in the datapath.
Gauge dp_flows_lookup_missed The number of incoming packets in a datapath not matching any existing flow in the datapath.
Gauge dp_flows_lookup_lost The number of incoming packets in a datapath destined for userspace process but subsequently dropped before reaching userspace.
Gauge dp_masks_hit The total number of masks visited for matching incoming packets.
Gauge dp_masks_total The number of masks in a datapath.
Gauge dp_masks_hit_ratio The average number of masks visited per packet. It is the ration between hit and total number of packets processed by a datapath.
Gauge interface Represents OVS interface. This is the primary metric for all other interface metrics. This metrics is always 1.
Gauge interface_admin_state The administrative state of the physical network link of OVS interface. The values are: down(0), up(1), other(2).
Gauge interface_link_state The state of the physical network link of OVS interface. The values are: down(0), up(1), other(2).
Gauge interface_mac_in_use The MAC address in use by OVS interface.
Gauge interface_mtu The currently configured MTU for OVS interface.
Gauge interface_of_port Represents the OpenFlow port ID associated with OVS interface.
Gauge interface_if_index Represents the interface index associated with OVS interface.
Gauge interface_tx_packets Represents the number of transmitted packets by OVS interface.
Gauge interface_tx_bytes Represents the number of transmitted bytes by OVS interface.
Gauge interface_rx_packets Represents the number of received packets by OVS interface.
Gauge interface_rx_bytes Represents the number of received bytes by OVS interface.
Gauge interface_rx_crc_err Represents the number of CRC errors for the packets received by OVS interface.
Gauge interface_rx_dropped Represents the number of input packets dropped by OVS interface.
Gauge interface_rx_errors Represents the total number of packets with errors received by OVS interface.
Gauge interface_rx_frame_err Represents the number of frame alignment errors on the packets received by OVS interface.
Gauge interface_rx_missed_err Represents the number of packets with RX missed received by OVS interface.
Gauge interface_rx_over_err Represents the number of packets with RX overrun received by OVS interface.
Gauge interface_tx_dropped Represents the number of output packets dropped by OVS interface.
Gauge interface_tx_errors Represents the total number of transmit errors by OVS interface.
Gauge interface_collisions Represents the number of collisions on OVS interface.

kube-ovn-pinger

Network quality related metrics:

Type Metric Description
Gauge pinger_ovs_up If the ovs on the node is up
Gauge pinger_ovs_down If the ovs on the node is down
Gauge pinger_ovn_controller_up If the ovn_controller on the node is up
Gauge pinger_ovn_controller_down If the ovn_controller on the node is down
Gauge pinger_inconsistent_port_binding The number of mismatch port bindings between ovs and ovn-sb
Gauge pinger_apiserver_healthy If the apiserver request is healthy on this node
Gauge pinger_apiserver_unhealthy If the apiserver request is unhealthy on this node
Histogram pinger_apiserver_latency_ms The latency ms histogram the node request apiserver
Gauge pinger_internal_dns_healthy If the internal dns request is unhealthy on this node
Gauge pinger_internal_dns_unhealthy If the internal dns request is unhealthy on this node
Histogram pinger_internal_dns_latency_ms The latency ms histogram the node request internal dns
Gauge pinger_external_dns_health If the external dns request is healthy on this node
Gauge pinger_external_dns_unhealthy If the external dns request is unhealthy on this node
Histogram pinger_external_dns_latency_ms The latency ms histogram the node request external dns
Histogram pinger_pod_ping_latency_ms The latency ms histogram for pod peer ping
Gauge pinger_pod_ping_lost_total The lost count for pod peer ping
Gauge pinger_pod_ping_count_total The total count for pod peer ping
Histogram pinger_node_ping_latency_ms The latency ms histogram for pod ping node
Gauge pinger_node_ping_lost_total The lost count for pod ping node
Gauge pinger_node_ping_count_total The total count for pod ping node
Histogram pinger_external_ping_latency_ms The latency ms histogram for pod ping external address
Gauge pinger_external_lost_total The lost count for pod ping external address

kube-ovn-controller

kube-ovn-controller status metrics:

Type Metric Description
Histogram rest_client_request_latency_seconds Request latency in seconds. Broken down by verb and URL
Counter rest_client_requests_total Number of HTTP requests, partitioned by status code, method, and host
Counter lists_total Total number of API lists done by the reflectors
Summary list_duration_seconds How long an API list takes to return and decode for the reflectors
Summary items_per_list How many items an API list returns to the reflectors
Counter watches_total Total number of API watches done by the reflectors
Counter short_watches_total Total number of short API watches done by the reflectors
Summary watch_duration_seconds How long an API watch takes to return and decode for the reflectors
Summary items_per_watch How many items an API watch returns to the reflectors
Gauge last_resource_version Last resource version seen for the reflectors
Histogram ovs_client_request_latency_milliseconds The latency histogram for ovs request
Gauge subnet_available_ip_count The available num of ip address in subnet
Gauge subnet_used_ip_count The used num of ip address in subnet

kube-ovn-cni

kube-ovn-cni status metrics:

Type Metric Description
Histogram cni_op_latency_seconds The latency seconds for cni operations
Counter cni_wait_address_seconds_total Latency that cni wait controller to assign an address
Counter cni_wait_connectivity_seconds_total Latency that cni wait address ready in overlay network
Counter cni_wait_route_seconds_total Latency that cni wait controller to add routed annotation to pod
Histogram rest_client_request_latency_seconds Request latency in seconds. Broken down by verb and URL
Counter rest_client_requests_total Number of HTTP requests, partitioned by status code, method, and host
Counter lists_total Total number of API lists done by the reflectors
Summary list_duration_seconds How long an API list takes to return and decode for the reflectors
Summary items_per_list How many items an API list returns to the reflectors
Counter watches_total Total number of API watches done by the reflectors
Counter short_watches_total Total number of short API watches done by the reflectors
Summary watch_duration_seconds How long an API watch takes to return and decode for the reflectors
Summary items_per_watch How many items an API watch returns to the reflectors
Gauge last_resource_version Last resource version seen for the reflectors
Histogram ovs_client_request_latency_milliseconds The latency histogram for ovs request

微信群 Slack Twitter Support


Last update: October 18, 2022
Created: June 30, 2022

Comments

Back to top